Device, method, and program for detecting injury of quadruped

Abstract

A method for detecting an injury of a quadruped includes: receiving measured data that is based on a gait of the quadruped and includes acceleration in a top-bottom direction and acceleration in a front-back direction; determining timings of putting legs of the quadruped on the ground based on the acceleration in the top-bottom direction, included in the received measured data, and determining whether or not each of the legs is injured based on values, corresponding to the determined timing of putting the legs on the ground, of the acceleration in the front-back direction.

What is claimed is: 
     
       1. A device for detecting an injury of a quadruped, comprising:
 a memory; 
 and a processor coupled to the memory and configured to execute a process including: 
 receiving measured data measured by an acceleration sensor attached to a front chest region of a quadruped, the measured data being based on a gait of walk of the quadruped and includes acceleration in a top-bottom direction and acceleration in a front-back direction, the measured data indicating a movement of the chest region of the quadruped; 
 first determining timings of putting legs of the quadruped on a ground based on the acceleration in the top-bottom direction, included in the received measured data; 
 and second determining whether or not each of the legs is Injured based on values, corresponding to the determined timings of putting the legs on the ground, of the acceleration in the front-back direction; 
 classifying the measured data into multiple groups for each of the legs based on the timings of putting legs of the quadruped on the ground and the acceleration in the front-back direction within time intervals between multiple time points when the quadruped puts a target leg on the ground and when the quadruped puts another leg on the ground immediately after the target leg on the ground; 
 and detecting an injury to the quadruped based on the values within more than one of the classified groups. 
 
     
     
       2. The device according to  claim 1 , wherein the measured data further includes an angular velocity about a yaw axis, wherein the first determining includes determining a left or a right of each of the legs based on the angular velocity about the yaw axis, determining front or back of each of the legs based on the acceleration in the top-bottom direction, and determining timings of putting the left and right back legs on the ground based on the timings of putting the left and right front legs on the ground, and wherein the second determining includes determining whether or not each of the legs is injured based on the values, corresponding to the determined timings of putting the legs on the ground, of the acceleration in the front-back direction. 
     
     
       3. The device according to  claim 2 , wherein the classifying classifies, into the multiple groups for each of the legs, areas of ranges of a graph of the acceleration in the front-back direction within time intervals between multiple time points when the quadruped puts the target leg on the ground and when the quadruped puts another leg on the ground immediately after the target leg on the ground;
 and the detecting the injury detects whether or not each of the legs is injured based on a number, classified into a group of values including a maximum value and another group of values including a minimum value, of times that the quadruped puts the legs on the ground. 
 
     
     
       4. The device according to  claim 3 ,
 wherein in the second determining, a lame leg is determined based on an increasing or decreasing trend in the acceleration in the front-back direction by referring to a determination table indicating relationships between lame legs and increasing or decreasing trends in the acceleration in the front-back direction represented by the numbers of times that the quadruped puts the legs on the ground. 
 
     
     
       5. The device according to  claim 2 ,
 wherein the 
 classifying classifies, into the multiple groups for each of the legs, the numbers of plotted values of a graph of the acceleration in the front-back direction within time intervals between multiple time points when the quadruped puts the target leg on the ground and when the quadruped puts another leg on the ground immediately after putting the target leg on the ground; and 
 the detecting the injury detects whether or not each of the legs is injured based on a number, classified into a group of values including a maximum value and another group of values including a minimum value, of times that the quadruped puts the legs on the ground. 
 
     
     
       6. A method for detecting an injury of a quadruped, performed by a computer, the method comprising:
 receiving measured data measured by an acceleration sensor attached to a front chest region of a Quadruped, the measured data being based on a gait of walk of the quadruped and includes acceleration in a top-bottom direction and acceleration in a front-back direction, the measured data indicating a movement of the chest region of the quadruped; 
 first determining timings of putting legs of the quadruped on a ground based on the acceleration in the top-bottom direction, included in the received measured data; 
 and second determining whether or not each of the legs is injured based on values, corresponding to the determined timings of putting the legs on the ground, of the acceleration in the front-back direction; 
 classifying the measured data into multiple groups for each of the legs based on the timings of putting legs of the quadruped on the ground and the acceleration in the front-back direction within time intervals between multiple time points when the quadruped puts a target leg on the ground and when the quadruped puts another leg on the ground immediately after the target leg on the ground; 
 and detecting an injury to the quadruped based on the values within more than one of the classified groups. 
 
     
     
       7. A non-transitory computer-readable storage medium storing a program for detecting an injury of a quadruped, the program causes a computer, when executed, to perform a process comprising:
 first determining timings of putting legs of a quadruped on a ground based on acceleration in the top-bottom direction, included in received measured data measured by an acceleration sensor attached to a front chest region of the quadruped, the measured data indicating a movement of the chest region of the quadruped; 
 and second determining whether or not each of the legs is injured based on values, corresponding to the determined timings of putting the legs on the ground, of acceleration in the front-back direction included in the received measured data; 
 classifying the measured data into multiple groups for each of the legs based on the timings of putting legs of the quadruped on the ground and the acceleration in the front-back direction within time intervals between multiple time points when the quadruped puts a target leg on the ground and when the quadruped puts another leg on the ground immediately after the target leg on the ground; 
 and detecting an injury to the quadruped based on the values within more than one of the classified groups.